john_ellis - have I understood correctly that TR6 would be better replaced with a 2N2219A? I think I have some of those TO 39 clip-on heat sinks somewhere....
Since ceramic pickups are no longer used, I’d remove the network attached to that input and attach a 10:1 attenuator and that would fix it.
I doubt. Yellow, purple, orange, twice.
The left 33N has been repleaced.
Attachments
"have I understood correctly that TR6 would be better replaced with a 2N2219A? I think I have some of those TO 39 clip-on heat sinks somewhere...."
Yes. Although the recent BC547 series appear to be rated at 500mW it is usually specified that it is soldered to an FR4 board, meaning that the wires should conduct some of the heat away. That usually stipulates a minimum board area of copper, but I didn't see that in the Fairchild version of the datasheet. TO-5/TO-39 cans could dissipate 800mW or 1W (dependent on type) in free air (at 25C).
So even though the "free air" thermal resistance of the BC and 2N2219A are similar (290K/W) the thermal ressitance to the case of the TO-39 is 50C/W, so with a heat clip should be better than 100K/W.
Which means if you use a 2N2219A and putheatclips on them they should stay cooler than the plastic devices, which should improve your quiescent current thermal stability.
I assume the board you have uses a 2N2905A for the PNP driver, so that too should have a heat clip. If that is a plastic type, swap to the 2N2905A!
Failing that you could use BD139-BD140 but you would need to joggle the leads to fit and use suitable TO-126 heat clips.
Yes. Although the recent BC547 series appear to be rated at 500mW it is usually specified that it is soldered to an FR4 board, meaning that the wires should conduct some of the heat away. That usually stipulates a minimum board area of copper, but I didn't see that in the Fairchild version of the datasheet. TO-5/TO-39 cans could dissipate 800mW or 1W (dependent on type) in free air (at 25C).
So even though the "free air" thermal resistance of the BC and 2N2219A are similar (290K/W) the thermal ressitance to the case of the TO-39 is 50C/W, so with a heat clip should be better than 100K/W.
Which means if you use a 2N2219A and putheatclips on them they should stay cooler than the plastic devices, which should improve your quiescent current thermal stability.
I assume the board you have uses a 2N2905A for the PNP driver, so that too should have a heat clip. If that is a plastic type, swap to the 2N2905A!
Failing that you could use BD139-BD140 but you would need to joggle the leads to fit and use suitable TO-126 heat clips.
I'll get 2N2219As to go with the existing 2N2905As and fit heat sinks, and I'll try using BD139s instead of BC237Bs for TR4 and mount them on the heat sink in between the output TIP41Cs. I'm not fixed on complete originality, but I don't really want to stray too far - I'm not going for a silk purse, just a fair representation of the original design.
I was given some B&C 12PE32 drivers a couple of days ago, so it will be interesting to see what can be done with them and this little amp - they are very efficient!
I was given some B&C 12PE32 drivers a couple of days ago, so it will be interesting to see what can be done with them and this little amp - they are very efficient!
I fitted the 2N2219As and changed TR4 to BD139 and moved them to the heat sink. Now the left channel fuse has blown, and there is -600 mV DC at the left speaker output and 1.2 v DC at the right. No idea what I have done wrong...
I have removed the left channel fuse in order to do one channel at a time. I removed the replacement transistors BD139 and 2N2219A and they tested ok. Put them back again and I now read -77 mV at the output. Connected MM source to Magnetic input and a speaker to the output and nothing, not even a faint hum. Tried with CD on Radio input and still nothing. There was always a bit of a thump from the speakers at turn-on (there was on my old one) but now nothing.
I have checked all the traces with a magnifying glass, and will check all connections again, remaining baffled for the moment....
I have checked all the traces with a magnifying glass, and will check all connections again, remaining baffled for the moment....
In it's original form have you set the bias current at the suggested level and checked the speaker output voltage? Might be a red herring but could indicate a problem with the basic DC biasing.
That is a disappointment. Did you check that the BD139 collector pad was insulated from the chassis? The photos suggested you used an insulator but drilling holes can leave swarf which annoyingly shorts things out.
Did you reset the quiescent current bias pot to maximum resistance? - that would almost certainly be different from the setting using the BC237 or what was there before.
When you have one channel powered what voltages do you get on the transistors?
Pity you did not find three colours of wires to use for the BD139- I take it you checked which was C, E?
There should be no DC at the output. Do you get a larger AC reading? This may be signs of oscillation. May have to put a 10nF or 100nF capacitor across the C,E of the BD139 on the board. Really, there should have been a capacitor across it in the first place, which is a potential problem when making mods.
Did you reset the quiescent current bias pot to maximum resistance? - that would almost certainly be different from the setting using the BC237 or what was there before.
When you have one channel powered what voltages do you get on the transistors?
Pity you did not find three colours of wires to use for the BD139- I take it you checked which was C, E?
There should be no DC at the output. Do you get a larger AC reading? This may be signs of oscillation. May have to put a 10nF or 100nF capacitor across the C,E of the BD139 on the board. Really, there should have been a capacitor across it in the first place, which is a potential problem when making mods.
Last edited:
I did clean the edges of the holes with a larger drill bit, so I don't think there was any scarf left. I also used the little plastic washers with a shoulder and plastic bolts and nuts, so there shouldn't be any contact. I removed the BD139s then put them back, but I didn't test for a short before doing that. There isn't one now, at least!
I left the bias pots at around 2/3, so around 600 ohms
I didn't have a third colour wire to hand, but I did check and re-check several times that I got C and E the right way round. The braids are actually short enough to follow the wires visually as well as a continuity test, and I checked and double-checked the connections before marking the circuit board.
I am getting around -400 mV DC across the output, and around 1.2 v AC between output + and - , and the same 1.2 v AC between + or - and chassis.
I look for some capacitors to put across the BD139
Thank you for all the help - I'd be lost without it!
I left the bias pots at around 2/3, so around 600 ohms
I didn't have a third colour wire to hand, but I did check and re-check several times that I got C and E the right way round. The braids are actually short enough to follow the wires visually as well as a continuity test, and I checked and double-checked the connections before marking the circuit board.
I am getting around -400 mV DC across the output, and around 1.2 v AC between output + and - , and the same 1.2 v AC between + or - and chassis.
I look for some capacitors to put across the BD139
Thank you for all the help - I'd be lost without it!
As I had a quasi circuit already in sim I ran a quick AC analysis using the components you showed in post #1 (1nF feedback cap, 47nF base bypass and a 100pF on the collector (to ground)on the VAS transistor ). There is a nasty spike (as I have always seen with improperly compensated amplifiers like this) at about 3MHz. It is actually worse if the capacitor on the base of the VAS is 100nF when the 100pF is removed. I assume you have the version you posted the circuit of (post #1) with 47nF and 100pF? The spike is controlled if the base capacitor is increased to 100nF but the 100pF is still required.
This compensation scheme was often used and Mullard published several designs with it in their 1970's era books. It needs all three capacitors (feedback, base, collector) to work.
Bailey showed a much better scheme which is surprising that this has not become more widely used, although that really applied to a complementary output stage (which if you are adventurous could be implemented using a TIP42A !)
My recommendation is that if this appears to be oscillation then if your capacitor is 47nF increase it to 100nF, if your circuit omits the 100pF, add it preferably with 100 or 220 ohm resistor in series. I would not recommend this as a hifi solution by any means, but it is one which should apply to your circuit if you want it to remain close to the original.
So first things first - what are the voltages and can you detect any signs of oscillation?
This compensation scheme was often used and Mullard published several designs with it in their 1970's era books. It needs all three capacitors (feedback, base, collector) to work.
Bailey showed a much better scheme which is surprising that this has not become more widely used, although that really applied to a complementary output stage (which if you are adventurous could be implemented using a TIP42A !)
My recommendation is that if this appears to be oscillation then if your capacitor is 47nF increase it to 100nF, if your circuit omits the 100pF, add it preferably with 100 or 220 ohm resistor in series. I would not recommend this as a hifi solution by any means, but it is one which should apply to your circuit if you want it to remain close to the original.
So first things first - what are the voltages and can you detect any signs of oscillation?
The standard classic Miller compensation seems to work better! Not perfect, but if, as your AC reading suggests, it is oscillation, then rather than any of the existing capacitors, you could remove the 1nF feedback, the 100nF (or 47nF) base shunt and reroute the 100pF from the collector of the VAS transistor to its base. That could be an interpretation of the shorted wiring in your original post, but with that 100pF Miller none of the other capacitors should have been required. But take this step by step. Try connecting the 100pF across the collector of the VAS to its base while leaving the 47nf or 100nf in place and see if that stops the strange output.
Thank you! I'll try that as soon as I can, but I'm busy for a couple of days, and there is GCSE revision to help with (and she probably knows more about Physics than I do...)
Are we still discussing the schematics in #1? Then I'm heavily scratching my head why there's DC at the output at all.
Best regards!
Best regards!
If anyone wants a play with a sim of the power amp...
All default models, click and run.
To the OP 🙂
If you are going to salvage this you need to switch into a fault finding mindset. You seem to have gone from a perfectly good working amp to something teetering on the edge of being scrapped.
I would suggest you rig up a DBT (dim bulb tester), put the amp back to its original design concept and then concentrate on bringing it back to working condition.
The midpoint voltage (at the amp output before the coupling cap tells you much about what is going on. It should be approx half the supply voltage.
All default models, click and run.
To the OP 🙂
If you are going to salvage this you need to switch into a fault finding mindset. You seem to have gone from a perfectly good working amp to something teetering on the edge of being scrapped.
I would suggest you rig up a DBT (dim bulb tester), put the amp back to its original design concept and then concentrate on bringing it back to working condition.
The midpoint voltage (at the amp output before the coupling cap tells you much about what is going on. It should be approx half the supply voltage.
Mooly - the OP complained that the quiescent current was not stable. It was a poor design not to have the bias transistor on the heatsink/chassis next to the OP transistors. Also, your simulated circuit has the 100pF Miller capacitor on the VAS transistor. With a short circuit in the diagram it could have been either. Having the Miller capacitor is a stable option- but a pretty harsh combination to have both a large base bypass and a large feedback capacitor. But that is exactly what I suggested in my last post, as the most likely explanation of voltages appearing at the output (assuming this is not hum pickup) is due to oscillation.
Could have been a much better design.
Could have been a much better design.